Skip to content

Code in PyTorch for the convex combination linear IAF and the Householder Flow, J.M. Tomczak & M. Welling

Notifications You must be signed in to change notification settings

jmtomczak/vae_vpflows

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

18 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

VAE with Volume-Preserving Flows

This is a PyTorch implementation of two volume-preserving flows as described in the following papers:

  • Tomczak, J. M., & Welling, M., Improving Variational Auto-Encoders using Householder Flow, arXiv preprint, 2016
  • Tomczak, J. M., & Welling, M., Improving Variational Auto-Encoders using convex combination linear Inverse Autoregressive Flow, arXiv preprint, 2017

Data

The experiments can be run on four datasets:

  • static MNIST: links to the datasets can found at link;
  • binary MNIST: the dataset is loaded from Keras;
  • OMNIGLOT: the dataset could be downloaded from link;
  • Caltech 101 Silhouettes: the dataset could be downloaded from link.

Run the experiment

  1. Set-up your experiment in experiment.py.
  2. Run experiment:
python experiment.py

Models

You can run a vanilla VAE, a VAE with the Householder Flow (HF) or the convex combination linear Inverse Autoregressive Flow (ccLinIAF) by setting model_name argument to either vae, vae_HF or vae_ccLinIAF, respectively. Setting number_combination for vae_ccLinIAF to 1 results in vae_linIAF.

Citation

Please cite our paper if you use this code in your research:

@article{TW:2017,
  title={{Improving Variational Auto-Encoders using convex combination linear Inverse Autoregressive Flow}},
  author={Tomczak, Jakub M and Welling, Max},
  journal={arXiv},
  year={2017}
}

Acknowledgments

The research conducted by Jakub M. Tomczak was funded by the European Commission within the Marie Skłodowska-Curie Individual Fellowship (Grant No. 702666, ”Deep learning and Bayesian inference for medical imaging”).

About

Code in PyTorch for the convex combination linear IAF and the Householder Flow, J.M. Tomczak & M. Welling

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages